1 of 30

Future Directions for Tropical Cyclone Research

Frank Marks

Retired, AOML/Hurricane Research Division

1

Marks – 11/05/2025

2 of 30

Frank circa 1968

Weatherwise, Vol.75, 1

2

Education:

Belknap College B.S. 1973
Massachusetts Institute of Technology M.S. 1975
Massachusetts Institute of Technology Sc.D. 1981

Professional Experience:

45 years at NOAA as a Research Meteorologist
Former director of NOAA/AOML/Hurricane Research Division
Former science lead of NOAA’s Hurricane Forecast Improvement Program

Professional Activities:

Member American Meteorological Society (AMS)
Member, American Association for the Advancement of Science (AAAS)
Member, Sigma Xi

Who Am I?

Marks – 11/05/2025

3 of 30

3

REDDIT AMA

Marks – 11/05/2025

4 of 30

72-hr Track Forecast

4

Current State of the Art

Operational Forecast Performance

Courtesy John Cangialosi & James Franklin (NWS/NHC)

72-hr Intensity Forecast

64% decrease

50% decrease

Marks – 11/05/2025

5 of 30

How Did We Get Here?

5

Unified approach to guide & accelerate forecast improvements since 2008

improve prediction of rapid intensification & track
improve forecasts & communication of storm hazards
incorporate risk communication research to create more effective products

http://www.hfip.org

Marks – 11/05/2025

Gall, R., J. Franklin, F. D. Marks, E. N. Rappaport, F. Toepfer, 2013: The Hurricane Forecast Improvement Project. Bull. Amer. Meteor. Soc., 94, 329–343 doi: https://dx.doi.org/10.1175/BAMS-D-12-00071.1.

Hurricane Forecast Improvement Program (HFIP) began in 2009 and has significantly improved forecast performance for the last 10 years, while meeting the 5-year goal to reduce track and intensity errors by 20 percent
HFIP provides the unifying organizational infrastructure and funding for NOAA and other agencies to coordinate the research needed to significantly improve guidance for TC track, intensity, and storm-surge forecasts, as well as accelerate the transition of R2O
HFIP focuses activities to research, develop, demonstrate, and implement enhanced operational modeling capabilities, dramatically improving the numerical forecast guidance made available to the National Hurricane Center.
HFIP facilitates the development of the next generation of tropical cyclone (TC) researchers for NOAA.
HFIP will build upon the original goals of the program through the following specific goals:

Reduce numerical forecast guidance errors, including during rapid intensification, by 50 percent from 2017;
Produce 7-day forecast guidance that is similar to the 2017 5-day forecast guidance;
Improve guidance on pre-formation disturbances, including genesis timing, and track and intensity forecasts, by 20 percent from 2017; and
Improve hazard guidance and risk communication, based on social and behavioral science, to modernize the TC product suite (i.e., products, information, and services) for actionable lead times for storm surge and all other threats. http://www.hfip.org/
Percent improvement is determined by evaluating track, intensity, storm size, and rapid intensification error relative to those over the 3-year period, 2015-2017

6 of 30

6

Characterizing & understanding these processes & their interactions are key steps in forecast improvement

Multiscale nature of processes are major reason for this difficulty

Challenge

Thunderstorm

(1 km)

Vortex

(100 km)

Turbulence

(0.001-0.1 km)

Environment

(1000 km)

Marks – 11/05/2025

7 of 30

How?

7

Observations

Modeling

Analysis systems

Evaluation

Initialization

Observing strategies

Understanding & Prediction of TCs

Impacts

Marks – 11/05/2025

8 of 30

Observations

8

https://www.aoml.noaa.gov/data-products/#hurricanedata

Courtesy Jason Dunion (CIMAS/HRD)

Marks – 11/05/2025

9 of 30

Modeling

Advance Hurricane Forecast Guidance: HWRF -> HAFS

9

HAFS-Basin: Multi-Moving Nests

Courtesy Bill Ramstrom (AOML/HRD)

Hurricane Humberto (08L) 00 UTC 27 September 2025

HWRF

HAFS

Marks – 11/05/2025

10 of 30

Data Assimilation

Improve Forecast Guidance

10

Utilize wealth of data from recon missions

Improve DA Science

Utilize unique data to advance TC DA science

ANALYSIS

OBSERVED RADAR

https://www.emc.ncep.noaa.gov/gc_wmb/vxt/HATCF/

Milton Observations Assimilated into HAFS

Hurricane Milton:

Marks – 11/05/2025

11 of 30

Emerging Research Theme

Advancing Emerging Technologies

11

- Small Uncrewed Aircraft Systems sUAS

Air-Deployed:

Black Swift S0 (1.5h, 2 lb)

Anduril Altius 600 (3-4h, 20 lb)

Land-Launched:

Black Swift S0 “hybrid” (1.5h)

Dragoon Coriolis (18h/1000 nmi range)

- Advanced Atmospheric Profilers

Skyfora Streamsondes: 8 at once

- Uncrewed Surface Vessel

Saildrone

- Uncrewed Ocean Profilers

Gliders

Marks – 11/05/2025

12 of 30

Emerging Research Theme

Machine Learning

12

for Hurricane Milton

µ = length (similar to normal distribution mean)

σ = scale (similar to normal distribution the stn. dev.)

γ = skewness

τ = tail

TCANE V1.0

Courtesy Mark DeMaria (CIRA)

https://deepmind.google.com/science/weatherlab

Marks – 11/05/2025

13 of 30

13

Observations

Modeling

Analysis systems

Improve Resiliency

Future Research Theme

Social, Behavioral, Economic Sciences

Climate Resilience to TC Impacts:

https://www.ncdc.noaa.gov/billions/

Marks – 11/05/2025

14 of 30

Low-Probability, High-Consequence Events

Risk: the potential of gaining or losing something of value

Risk = Probability × Consequence × Vulnerability

Risk perception: the subjective judgment people make about the severity & probability of a risk, which varies from person to person

Actual Risk ≠ Perceived Risk

14

Probability

Consequence

Probability

Consequence

Low Risk

High Risk

Vulnerability

Marks – 11/05/2025

15 of 30

U.S. Tropical Cyclone Risk

15

1980-2024

https://www.ncdc.noaa.gov/billions/

Marks – 11/05/2025

16 of 30

Future Challenges: Impacts

Not Dependent on Saffir-Simpson Category

Marks – 11/05/2025

17 of 30

Future Challenges

Storm Surge:

Track, Intensity, & Size
Accurate digital elevation and bathymetric data
Tides, waves, thermohaline steric affects
Coupled inland-coastal flooding forecast guidance
Unite hydrologic regime in coastal watershed with the coastal ocean

Marks – 11/05/2025

18 of 30

Future Challenges

Precipitation/Flooding:

Terrain
Land use
Antecedent rainfall
Soil Moisture
Seasonal changes
Catchment basin

2024 Atlantic Season U.S. Direct Fatalities

Marks – 11/05/2025

19 of 30

Future Challenges

Wind & Severe Weather:

Gust factor
Tornados
Terrain/land use effects
Soil Moisture
Seasonal changes (foliage)

Time series of wind speed, Hurricane Harvey

Fernández-Cabán et al. (2019, BAMS)

Theoretical GF

Marks – 11/05/2025

20 of 30

Some Key Science questions/ideas to explore:

Tornado related questions:

What is PDF of hurricane produced tornado gust factors (ratio of tornado wind/background mean wind)?
What is PDF of tornado duration?
What is relationship between size of a TC & number/strength of tornadoes generated?
What is relationship between environmental (low-level) CAPE/helicity etc. & tornado genesis.

Wind gust factor related questions:

What is PDF of gust factors vs radial distance, distance inland, terrain, etc.
What is relationship between gust factors & convective features from radar imagery
What is relationship between vertical wind profile at a particular location & observed gust factor, particularly as a function of distance from the coastline.

Wind & Severe Weather:

Courtesy John Kaplan (AOML/HRD)

Marks – 11/05/2025

21 of 30

Stakeholder needs: Federal, state & local climate risk assessment & resiliency

Storm surge & inland flooding uncertainty, catastrophe risk models, insurance, & reinsurance industries
Storm-scale high-resolution reanalysis within global reanalysis to train AI/ML models

NOAA research capability available

Expertise in Seasonal to Sub-Seasonal research & guidance
Expertise in both data, science, modeling, & DA
High-resolution modeling & DA system
Data (>40 years of TC, ocean, & satellite observations)

21

Future Challenges

Courtesy Xuejin Zhang (AOML/HRD)

Marks – 11/05/2025

22 of 30

Interdisciplinary partnerships & collaborations

Focus on local & regional climate adaptation & resiliency (e.g., U.S. Climate Resiliency Toolkit, University of Miami Resiliency Academy)
Bring physical science research expertise to bear on issues related to tropical cyclone resiliency

22

Future Research Theme

Marks – 11/05/2025

23 of 30

Example: Risk Modeling

State of Florida Public Hurricane Loss Model

Modeled 50,000 years of hurricane activity
Winds input to damage model & losses aggregated
Average annual loss computed at each zip code

23

Zero Deductible Loss Cost by Zip Code for Owners Frame

Courtesy Bachir Annane (AOML/CIMAS)

Main Components of the FPHLM

Marks – 11/05/2025

24 of 30

Questions?

24

https://vlab.noaa.gov/web/osti-modeling/hfip/

https://www.aoml.noaa.gov/hurricane-research-division/

frank.marks@noaa.gov

AOML

FACETs

Marks – 11/05/2025

25 of 30

25

NOAA student opportunities

Marks – 11/05/2025

26 of 30

26

G-IV

P-3

NOAA Hurricane Hunters

https://www.omao.noaa.gov/learn/aircraft-operations

Marks – 11/05/2025

27 of 30

Observations

27

https://www.aoml.noaa.gov/hrd/Storm_pages/dorian2019/mission.html

https://www.aoml.noaa.gov/dynamics-and-physics/

Hurricane Dorian at ~1324 UTC 1 September 2019

Courtesy Michael Fischer (HRD/CIMAS)

Marks – 11/05/2025

28 of 30

28

10 mi

Hurricanes FAQ

Marks – 11/05/2025

29 of 30

Emerging Research Theme

Impacts

29

Supporting transitions at the Hurricane & Ocean Testbed (HOT)
Working with social scientists
Sustain partnerships with NWS & others

William Lapenta Lab at NHC

Marks – 11/05/2025

30 of 30

30

Emerging Research Theme

Courtesy Castle Williamsberg (OAR/WPO)

Big Themes & Takeaways from Triangulation Efforts

Identify ways to localize & personalize TC information

Improve the accessibility of TC products and services.

People search for different types of information during different phases of the lifecycle of a TC threat.

Timing information is critical for decision-making, thus timing of when forecasts are issued is important too.

Uncertainty information is important to communicate,

but it is not always communicated well.

Graphical TC products are important, but some need to improve their depiction of risk and/or uncertainty.

There is a misperception among forecasters & partners that the public does not understand uncertainty info.

There is a misperception that emergency managers are highly numerate like

weather forecasters.

Improve forecast communication of hazards

Marks – 11/05/2025